Glad you like it!
This effect comes from that split between rendering and presentation stages again. Just to explain for everyone why this works.
Mode 13h was a really popular VGA display mode back in the day, with a 320x200 16:10 resolution target. But the most common monitors were all 4:3 aspect for presentation, so the target render ended up being displayed 20% higher in output signal. As you already know, game developers worked around this by manually drawing the art a bit squishy so it came out OK on the target hardware. Great then, but not so fun to deal with on modern hardware.
Previously DFU was both rendering and presenting at the set display resolution. Now with retro rendering enabled, its rendering into a true 320x200 or 640x400 resolution target then presenting to whatever your set game resolution is. As a final step, this is then scaled to your desktop resolution with aspect correction resulting in letterbox/pillarbox bars.
The reason 1600x1200 works so well is that it's a 4:3 aspect that's exactly 5x wider and 6x higher than 320x200, so a 16:10 render is being presented as 4:3, which is essentially reproducing the intended display conditions.
A 5x width and 6x height multiplier is so important because 6 is precisely 20% more than 5. As 5 is a prime number that yields a perfect integer at 20%, there aren't many resolutions that will provide this same effect so exactly. There is no smaller resolution than 1600x1200 than can provide the exact same shape at presentation time with a perfect 20% increase to height.
It took me a while to really grok all of this.